<html>
<head>
<meta content="text/html; charset=UTF-8" http-equiv="Content-Type">
</head>
<body bgcolor="#FFFFFF" text="#000000">
<div class="moz-cite-prefix"><br>
<pre class="moz-signature" cols="0">
</pre>
Martin Millnert wrote on 28/02/2013 11:06:<br>
</div>
<blockquote cite="mid:1362042384.12712.4.camel@galileo.millnert.se"
type="cite">
<pre wrap="">Tassos,
On Thu, 2013-02-28 at 09:10 +0200, Tassos Chatzithomaoglou wrote:
</pre>
<blockquote type="cite">
<pre wrap="">Either their own (largest percentage), or ours (after they manually enable IPv6).
</pre>
</blockquote>
<pre wrap="">
Well there's your problem.
Do you have any statistics on how many have manually enabled IPv6 in
yours?
Or, better yet, could you map how many of the subscribers actually send
IPv6 packets at all?
You could use this to re-factor your traffic statistics to see what the
stats are per active dual-stack subscriber. Then your data should be
more up to par.
Regards,
Martin
</pre>
</blockquote>
These statistics were based on radius accounting, so unless there is
a bug on this bras, i don't any possibility to get counters for ipv6
packets without having actual ipv6 connectivity.<br>
<br>
<br>
<br>
<div class="moz-cite-prefix">Mike Jones wrote on 28/02/2013 18:00:<br>
</div>
<blockquote
cite="mid:CAAAas8FcR=0tRE3QKjJUMkgzs_G=Ns95_Tb=EN6Y-fGXNKW_PA@mail.gmail.com"
type="cite">
<div class="moz-text-plain" wrap="true" graphical-quote="true"
style="font-family: -moz-fixed; font-size: 13px;"
lang="x-unicode">
<pre wrap=""><span class="moz-txt-citetags">>></span>
</pre>
<blockquote type="cite" style="font-style: italic; color:
#33CC00;">
<pre wrap=""><span class="moz-txt-citetags">> </span>Can you be more specific with the term %age?
</pre>
</blockquote>
<pre wrap="">I know customer numbers are normally carefully released, but I expect
many people wouldn't mind sharing that out of 50,000 dual stack lines
sampled 10,000* did some kind of v6. I don't believe I have seen v4/v6
usage figures presented like this before.
I assume ISPs who provide their own CPE would see some v6 traffic from
</pre>
<blockquote type="cite" style="font-style: italic; color:
#33CC00;">
<pre wrap=""><span class="moz-txt-citetags">></span>95% of customers, but that might also be an interesting 'problem'
</pre>
</blockquote>
<pre wrap="">area if anyone has looked in to it.
- Mike
*wishful thinking?
</pre>
</div>
</blockquote>
<br>
After checking data for 1 day, we seem to have a ratio of 4:1
("dual-stack in wan" subscribers vs "ipv6_bytes>0" subscribers).<br>
<br>
<br>
<div class="moz-cite-prefix">Eric Vyncke (evyncke) wrote on
28/02/2013 21:30:<br>
</div>
<blockquote
cite="mid:A75A7963-2C20-44DF-BB78-F7252813079F@cisco.com"
type="cite">
<pre wrap="">The difference between down and up, can we attribute it to ver asymmetric flows such as YouTube?
Interesting data point, thanks
Eric
</pre>
</blockquote>
<br>
Most probably yes. I'm estimating that around 70% of downstream IPv6
traffic is from Google services.<br>
<br>
<br>
<br>
Just had a quick look at one of our dual-stack dns servers too
(being auto-provisioned to IPv6 subscribers only):.<br>
<tt><br>
</tt><tt>++ Incoming Queries ++</tt><tt><br>
</tt><tt> 33831803 A</tt><tt><br>
</tt><tt> 12383371 AAAA</tt><tt><br>
</tt><tt><br>
</tt><tt>++ Outgoing Queries ++</tt><tt><br>
</tt><tt> 14793131 A</tt><tt><br>
</tt><tt> 4745494 AAAA</tt><tt><br>
</tt><tt><br>
</tt><tt>++ Name Server Statistics ++</tt><tt><br>
</tt><tt> 31507 IPv4 requests received</tt><tt><br>
</tt><tt> 47212503 IPv6 requests received</tt><tt><br>
</tt><tt><br>
</tt><tt>++ Resolver Statistics ++</tt><tt><br>
</tt><tt> 20568450 IPv4 queries sent</tt><tt><br>
</tt><tt> 1947656 IPv6 queries sent</tt><tt><br>
</tt><tt> 19462258 IPv4 responses received</tt><tt><br>
</tt><tt> 1912116 IPv6 responses received</tt><tt><br>
</tt><tt> 558329 IPv4 NS address fetches</tt><tt><br>
</tt><tt> 575541 IPv6 NS address fetches</tt><tt><br>
</tt><tt> 17742 IPv4 NS address fetch failed</tt><tt><br>
</tt><tt> 529662 IPv6 NS address fetch failed</tt><tt><br>
</tt><tt><br>
</tt><tt>++ Cache DB RRsets ++</tt><tt><br>
</tt><tt> 75761 A</tt><tt><br>
</tt><tt> 3260 AAAA</tt><tt><br>
</tt><tt> 45 !A</tt><tt><br>
</tt><tt> 8407 !AAAA</tt><br>
<br>
<br>
<br>
--<br>
Tassos<br>
<br>
<br>
</body>
</html>